Pandas 用相应列的值填充多个列，而不对每个列重复

Question

假设我有一个这样的 DataFrame：

x = pd.DataFrame({'col1_x': [15, np.nan, 136, 93, 743, np.nan, np.nan, 91] ,                                        
                  'col2_x': [np.nan, np.nan, 51, 22, 38, np.nan, 72, np.nan], 
                  'col1_y': [10, 20, 30, 40, 50, 60, 70, 80],
                  'col2_y': [93, 24, 52, 246, 142, 53, 94, 2]})

我想分别用col_y中的值填充col_x中的NaN值，我可以这样做：

x['col1_x'] = x['col1_x'].fillna(x['col1_y'])
x['col2_x'] = x['col2_x'].fillna(x['col2_y'])

print(x)

将产生：

   col1_x  col2_x  col1_y  col2_y
0    15.0    93.0      10      93
1    20.0    24.0      20      24
2   136.0    51.0      30      52
3    93.0    22.0      40     246
4   743.0    38.0      50     142
5    60.0    53.0      60      53
6    70.0    72.0      70      94
7    91.0     2.0      80       2

但是需要用不同的变量重复相同的函数，现在让我们假设我有一个更大的 DataFrame 和更多的列，是否可以不重复就可以做到？

Answer 1

您可以使用以下符号 -

x.fillna({"col1_x": x["col1_y"], "col2_x": x["col2_y"]})

假设你可以提取所有你能做的索引数字 -

replace_dict = {f"col{item}_x":x[f"col{item}_y"] for item in indices}
x = x.fillna(replace_dict}

Answer 2

您是否正在尝试制作此类功能：

def fil(fill,fromm):
    fill.fillna(fromm,inplace=True)

fil(x['col1_x'],x['col1_y'])

或者，如果您对 dataframe(x) 有把握，那么：

def fil(fill,fromm):
    x[fill].fillna(x[fromm],inplace=True)

fil('col1_x','col1_y')

对于您的代码：

import pandas as pd
import numpy as np

x = pd.DataFrame({'col1_x': [15, np.nan, 136, 93, 743, np.nan, np.nan, 91] ,                                
                  'col2_x': [np.nan, np.nan, 51, 22, 38, np.nan, 72, np.nan], 
                  'col1_y': [10, 20, 30, 40, 50, 60, 70, 80],
                  'col2_y': [93, 24, 52, 246, 142, 53, 94, 2]})


def fil(fill,fromm):
    x[fill].fillna(x[fromm],inplace=True)

fil('col1_x','col1_y')
fil('col2_x','col2_y')

print(x)

"""
   col1_x  col2_x  col1_y  col2_y
0    15.0    93.0      10      93
1    20.0    24.0      20      24
2   136.0    51.0      30      52
3    93.0    22.0      40     246
4   743.0    38.0      50     142
5    60.0    53.0      60      53
6    70.0    72.0      70      94
7    91.0     2.0      80       2
"""

此外，如果您有像 col1_x、col2_x、col3_x 这样的列名称...对于 y 也是如此，那么您可以像这样自动执行它：

for i in range(1,3):
    fil(f'col{i}_x',f'col{i}_y')

Answer 3

你可以使用 **kwargs 到 assign()
构建一个 dict 并理解构建 **kwargs

import pandas as pd
import numpy as np
x = pd.DataFrame({'col1_x': [15, np.nan, 136, 93, 743, np.nan, np.nan, 91] ,                                        
                  'col2_x': [np.nan, np.nan, 51, 22, 38, np.nan, 72, np.nan], 
                  'col1_y': [10, 20, 30, 40, 50, 60, 70, 80],
                  'col2_y': [93, 24, 52, 246, 142, 53, 94, 2]})

x.assign(**{c:x[c].fillna(x[c.replace("_x","_y")]) for c in x.columns if "_x" in c})

	col1_x	col2_x	col1_y	col2_y
0	15	93	10	93
1	20	24	20	24
2	136	51	30	52
3	93	22	40	246
4	743	38	50	142
5	60	53	60	53
6	70	72	70	94
7	91	2	80	2

它是如何工作的

# core - loop through columns that end with _x and generate it's pair column _y
{c:c.replace("_x","_y") 
 for c in x.columns if "_x" in c}

# now we have all the pairs of a columns let's do what we want - fillna()
{c:x[c].fillna(x[c.replace("_x","_y")]) for c in x.columns if "_x" in c}

# this dictionary matches this function.... https://pandas.pydata.org/docs/reference/api/pandas.DataFrame.assign.html
# so final part is call the function with **kwargs
x.assign(**{c:x[c].fillna(x[c.replace("_x","_y")]) 
            for c in x.columns if "_x" in c})

Pandas 用相应列的值填充多个列，而不对每个列重复

Pandas fillna multiple columns with values from corresponding columns without repeating for each

python

dataframe

python-3.x

pandas

fillna

它是如何工作的

	col1_x	col2_x	col1_y	col2_y
0	15	93	10	93
1	20	24	20	24
2	136	51	30	52
3	93	22	40	246
4	743	38	50	142
5	60	53	60	53
6	70	72	70	94
7	91	2	80	2

	col1_x	col2_x	col1_y	col2_y
0	15	93	10	93
1	20	24	20	24
2	136	51	30	52
3	93	22	40	246
4	743	38	50	142
5	60	53	60	53
6	70	72	70	94
7	91	2	80	2

	col1_x	col2_x	col1_y	col2_y
0	15	93	10	93
1	20	24	20	24
2	136	51	30	52
3	93	22	40	246
4	743	38	50	142
5	60	53	60	53
6	70	72	70	94
7	91	2	80	2